Closed JonasBa closed 1 year ago
@JonasBa Thank you for your contribution!
I'd like to merge #76 first and ask you to rebase/merge those changes afterwards as both impact lib/poparser.js
.
I'll leave a comment when that's done.
@JonasBa The PR I mentioned earlier has been merged. Would you be able to rebase your work on the updated master branch?
@smhg sorry about that, the GH notification got buried to 2nd page... I rebased the PR to the re-synced fork, it should be ok to review now.
@JonasBa thank you! Looking at the changes, I would think returning early in the loops makes all the difference? The loop at the end I'd rather keep as it gets a bit verbose now. Are you ok if I revert that part? And I also think it would be nice to define a key-to-regex Map object so we can loop over that. But that's a separate refactor.
@JonasBa I've published a 7.0.1 version which returns early in the outer loop. Could you please try that in your project(s)? This might be the main difference. Other refactors are welcome too, but let's have a look at this first.
@smhg thanks so much, your first reply got buried in my notifications and I only saw it after cleaning up my inbox (result of coming back from vacation). Thanks for making the changes and sorry I missed your reply
First of all, thanks for the great project, we use it at Sentry via our po-catalog-loader.
We recently started profiling our webpack build processes and noticed that compared to regular parsing of the files, a significant amount of time was spent in the _parseComments fn (all of the red colored frames are calls to _parseComments fn)
At a quick glance, I suspect that this is due to the amount of string allocations being made - full transparency, I have not benchmarked any of this but would be happy to it if it makes for a stronger case for the changes in this PR.
changes: